Choose a skylight cleaning robot by the highest glass edge you must reach: under 12 m suits a battery machine, 12-25 m suits a tether-plus-battery hybrid, and above 25 m needs a tethered unit with fall-arrest anchoring. Lingdu Intelligence splits this into the Lingyun Y3 and Lingfeng S1 for mid roofs, and the Lingkong K3 for taller atria.
Height is the number that quietly rules the rest of the spec sheet. Get it wrong and you either pay for reach you never use, or you buy a machine that cannot safely cross the deepest bay of your atrium.
Why does roof height change the robot you buy?
A robot working at 8 m and a robot at 30 m face different problems. At 8 m the crew can see the machine, water can be carried onboard, and a mistake is annoying. At 30 m you are relying on tether tension, fall-arrest lines, and remote diagnostics because nobody is climbing up to nudge it.
That gap shows up in three places: the power system, the anchoring method, and the price. A 12 m glass roof over a shopping mall is a very different job from the 42 m atrium of a commercial tower.
| Roof height | Typical power | Anchor needs | Rough price band (2026) |
|---|---|---|---|
| Up to 12 m | Onboard battery, 4-6 h | Low guard rail or edge stop | USD 12,000-20,000 |
| 12-25 m | Hybrid tether + battery | Edge anchor points, 2 per bay | USD 22,000-34,000 |
| Above 25 m | Tethered, mains power | Certified fall-arrest anchor | USD 35,000-60,000 |
What suction strength do you actually need?
Suction keeps the robot stuck to glass, and the number that matters is holding force per kilogram of machine. For near-vertical curtain wall you want the highest figure you can afford, because gravity is pulling sideways. On a shallow skylight slope, 15 degrees or less, the load is gentler.
A workable floor is around 2200 Pa for horizontal or low-slope glass, and 3000 Pa or more for anything over 45 degrees. Below that, dust and dried water spots start breaking the seal. This is also why wet-cleaned glass is safer than dry glass: the film helps the cup seal.
Which roofs should not get a robot?
Blunt answer: don’t automate a roof that is cleaned twice a year. The payback math needs regular cycles. A small office building with a 30 m2 skylight cleaned on a casual schedule will never justify USD 25,000 of hardware plus training.
Also skip robots on fragile single-glazing older than the 1990s, badly crazed sealant, or frames that flex under load. I have seen a frame move 4 mm under a test weight, and that is a stop signal. Get a structural check before you buy, not after.
What does the 2026 price look like in practice?
The Lingyun Y3 sits in the mid band, aimed at 15-25 m commercial atria where a hybrid power line is practical. The Lingfeng S1 handles lower slopes and tighter bays where a lighter body wins. The Lingkong K3 covers taller, more exposed roofs where tether safety and long reach matter most.
Budget another 8-12% for the first year: spare cups, brush sets, a filter kit, and two days of operator training. Buyers who skip this tend to call the supplier in month four with a machine that will not hold a seal.
How do you test before signing?
Ask for a trial on your worst bay, not the easiest one. The worst bay is the one with the deepest valley, the steepest pitch, and the dirtiest glass. If the robot handles that in under 20 minutes per cycle, the rest follows.
Talk to the team at Skylight Cleaning Robot about your roof drawings before you commit to a model. A 15-minute call about bay depth and slope saves weeks of retrofit pain.
Key Takeaways
- Height first: battery under 12 m, hybrid 12-25 m, tethered above 25 m.
- Suction floor of about 2200 Pa for flat glass, 3000 Pa for steep pitches.
- Low-use small skylights rarely justify a robot – regular cycles do.
- Budget 8-12% of purchase price for first-year consumables and training.
- Test on your worst bay, never the easiest one.
Does roof shape matter more than total area?
Yes, and this trips up a lot of buyers. Two roofs of identical size can differ by a factor of three in how long a machine takes. A single clear rectangle is the easy case. An L-shaped atrium with a glass strip around a planted side, or a sawtooth factory roof, forces the robot to re-anchor repeatedly and eats throughput.
Walk your roof and count the frame breaks before you talk to any supplier. Bring that number to the call. A vendor quoting 500 m2/h who hears “40 mullions across the run” will either revise the number or admit the machine is wrong for the roof. Both answers are useful.
What should the commissioning day include?
Set aside a full day, run a dry cycle first, then a wet one. Check that the tether length covers the furthest bay with slack to spare, verify the anchor torque, and record a baseline vacuum reading for each cup. Those baseline numbers are what you compare against in month six when something starts to feel off.
Train two people, not one. If the single trained operator is on leave, you do not want a machine sitting idle or, worse, being driven by someone guessing.